Airtable counts: linked records versus populated values

SOURCE-BASED ANALYSIS · NO HANDS-ON RATING

Name the unit you count

Airtable’s Count field counts records in a designated linked record field. Its documentation also describes rollup approaches: COUNTALL(values) includes linked records with empty selected values, while COUNTA(values) counts non-empty values. These answer different questions. A result is meaningful only when the reader knows its unit and population.

Example: attendance and completed feedback

Imagine one workshop linked to four attendee records. Three have a filled feedback field and one has not replied. “Attendees linked” should be four; “feedback responses present” should be three. Neither number tells you whether a response was thoughtful or an attendee actually joined. Keep those business definitions separate.

Create a fixture with an intentional blank

1. Create the four synthetic attendees and link each to the workshop.
2. Add a Count field over the Attendees link and label it Attendees linked.
3. For response presence, configure an appropriate rollup over the feedback field using COUNTA(values).
4. Leave one feedback value empty and check the expected four-versus-three result.
5. Remove one known link in the disposable fixture and confirm that you can explain how each measure changes.

Avoid denominator errors

Under these assumptions, three responses out of four linked attendees gives a 75% response rate. Label it as a hypothetical calculation. If cancelled attendees remain linked, the denominator may need a separate eligibility condition. If one attendee has several feedback rows, counting those rows can overstate the number of people responding. Decide whether the unit is a person, a submission or a booking before reporting a rate.

When a simple count is enough

Use the direct Count field when the relationship count is exactly the question. Use a more specific calculation when blank values or eligibility conditions change the meaning. Maintain one sample with no links and one with an incomplete response so later schema edits cannot silently change what your dashboard promises. This is a suggested review practice, not a measured validation of any live workspace.

Continue with Airtable rollup totals: why a filtered view does not change the sum when that next step matches your workflow.